Experimental partitioning of Sr and Ba in Kiglapait feldspars by S.A. Morse and J. Allaz (American Mineralogist, November-December, vol. 98, p. 2197–2200, 2013, Article DOI: http://dx.doi.org/10.2138/am.2013.4630. Addendum DOI: http://dx.doi.org/10.2138/am.2014.641).
This Addendum addresses the issue of retrieving liquid Sr compositions from the Sr content of natural feldspars. A partition coefficient defines the equilibrium between two phases. In igneous petrology the partitioning between crystals and liquid at the liquidus may be used to calculate fractionation during crystallization. The high values for strontium partitioning between feldspar and liquid in our paper are well suited to that purpose, but we found that the feldspar compositions would not suit for inverting them to liquid compositions. They gave results that were too rich in Sr, as shown in our Figure 4. We attributed that to an excess of Sr captured by the feldspars from the rocks. This idea was unnecessary. We show here the dilution effect that occurs with crystal growth, and find another route to inverting Sr in feldspar to liquid concentrations by simply using the natural feldspars compared to the experimental melts.
The experimental feldspars contain large amounts of Sr in small amounts of experimental feldspar, 1–60% crystals as complementary to the “% glass” listed in Table 1 of the original paper. Indeed, all but two of the entries have ≤30% feldspar. Mass balance is preserved when crystals grow in a closed system. If these crystals grew to nominal fractions of the whole rock, ranging from 75% at the start of crystallization to 63% at the end, their Sr content would be distributed over the larger mass of the crystals and would approximate the bulk composition of Sr in the rocks.
This proposition is easily tested in principle by simply dividing the experimental feldspar compositions by the fraction of oxygen-normative feldspar in the Kiglapait rocks, as modeled from whole-rock compositions. When this is done (Addendum Table 1 here), the diluted Sr contents of the bulk feldspars plot among or below the natural feldspars in Figure 1 here. The reductions are a bit excessive, but the principle is clear. The experimental feldspars have not scavenged Sr from other components but are rich in Sr simply because they are small in volume. Their partition coefficients (original Fig. 2) represent what one would use to describe the infinitesimal effect of plagioclase fractionation at the liquidus.
In practical application, the feldspar compositions can be inverted to liquid compositions by use of a reduced partition coefficient as suggested by Figure 1 and derived in Addendum Table 1 here. For this purpose the least loss of information will result if the natural feldspar compositions are simply ratioed to the experimental glass compositions. To do this, we describe the experimental glass compositions as a function of the An content of the crystals (listed in Morse 1982 and in Deposit Item Table 11 here). This is done in two equations:
Then the partition coefficients are formed as usual, D = Sr(fsp)/Sr(gl). These in turn are scaled to the plagioclase composition as shown in Figure 2 here. The results are best described in two equations as shown in the figure. As can be appreciated in Figure 1, at high An the natural feldspars are close to the glass compositions, yielding reduced D values near 1.2 in Fig. 2 instead of the full values near 1.8 of the original paper. They converge to 1.9 at An 30 and again rise steeply to D = 5.5 at An 12. By this exercise the natural plagioclase compositions in Sr can be inverted to liquid compositions, while the experimental values remain valid for fractionation at the liquidus.